search

Audio-WestlakeU / ATST-SED

This repo includes the official implementations of "Fine-tune the pretrained ATST model for sound event detection".
MIT License
93 stars 13 forks source link

Ask help about annotation files! #2

Closed Harper812 closed 1 year ago

Harper812 commented 1 year ago

Hi, thanks for your opening source. I don't to konw where to get files (strong_val_tsv and strong_val_dur). Would you please send to me via E-mail hby@bupt.edu.cn or explain how to generate these two files?

SaoYear commented 1 year ago

The strong_val_tsv and strong_val_dur files are just the splited train/validation sets from the strong_real set (3000+ clips) with the split ratio 9:1.

It actually turns out that this real validation set plays a trivial role on the model selection part.

So there is no need for you to use them. If you really want to apply them, you can use the attached files. But please make sure this your trian tsv file does not contain any of the validation files. audioset_strong_val.zip

Harper812 commented 1 year ago

Thank you for replying me instantly! Hope the scientific research goes well！

SaoYear commented 1 year ago

An update of the strong real data meta files, including: strong_real_train.tsv strong_real_val.tsv strong_real_val_duration.tsv audioset_strong_train_val.zip